5
9
0
The second digit had to be looked up in the leftmost column. It made sense that this column ran from 1 to 5: as Olson had noticed in his early analysis of the trinomes, the second digit in them was never 0 and never greater than 5. Similarly, the third digit of the trinomes was always either 1 or 2. So if the last two digits of a trinome were, say, 41—the single digit it would convert to could be found in the cell where the row corresponding to 4 intersects with the column under 1. That digit, as you can see from the conversion table, is 7.
As an example of the entire conversion from three-digit number to two-digit number, let’s consider the first of the 292 trinomes in the message that Regan marked as Letter M-I134. That trinome is 132. Following the rule laid out for conversion of the first digit, 1 becomes 0 (as per the first table). The second and third digits, 32, converted using the second table, become 6. So the trinome 132 is now reduced to 06.
Olson stripped off this first layer of encryption and converted the multiple lines of trinomes representing the three “letters” into a series of two-digit numbers. The second layer of encryption involved the three NRO phone lists that Regan said were his key. Both Olson and the code breakers at the NSA had suspected that Regan might have used a book code of some kind. Olson was pleased to learn that at least this conjecture had been valid. In Regan’s scheme, the phone lists essentially served the same purpose as a book.
Using the phone lists, every pair of numbers in the two-digit series derived from the trinomes could be converted into a letter or a digit. (This explained another feature of the letters that Olson had noted earlier: the fact that each constituted an even number of trinomes.) The first two-digit number indicated how many names down on the list you had to go: if the number was 12, say, you had to go down to the twelfth name. The second two-digit number gave the position of the character from that particular listing that the pair represented.
Letter M-I134 began with the following trinomes:
132 111 132 011 012 112 021 142 042 132 042 011 752 121 . . .
Stripping the first layer of encryption translated these to:
06 01 06 11 12 02 13 08 15 08 18 06 18 11 . . .
Each pair in this series stood for a letter or a digit. Demarcating those pairs from one another made the series look something like this:
(06 01) (06 11) (12 02) (13 08) (15 08) (18 06) (18 11) . . .
Now, to decode the first pair of numbers (06 01), the reader of the message had to go down to the sixth name on the particular phone list that was meant to be used as the key. That name happened to be “Saxby, Robert.” The first character in the listing was S—and so the pair (06 01) became S.
Converting the rest of the series resolved the opening of the message to: “S T A R T L Y,” which was Regan’s way of denoting that the coordinates to follow were for a package intended for Libya. Then came the coordinates themselves, encrypted in their entirety: N3720253 and W7734312. (Since each listing contained a name as well as a phone number, Regan could use the list to encrypt both letters and digits.) Read in the “degrees decimal minutes” format, the coordinates represented a latitude of 37 degrees 20.253 minutes and a longitude of 77 degrees 34.312 minutes. Going beyond the coordinates, the message read “FT 7,” marking the number of feet to the tree on which the nails had been hammered. This was followed by the word “END.” The words “START” and “END” were like bookends—linguistic markers bracketing each set of coordinates.
Olson would eventually decrypt all three letters in this way. But when he came in to work on the trinomes to help find the twelfth package, Carr and his fellow agents still had not gotten ahold of the three phone lists. Despite that handicap, Olson believed he could figure out the desired coordinates.
The reason was that Regan had built into his scheme what cryptanalysts call a “back door”—a way to recover the information in case he lost the key. As with numerous elements of his plot, here, too, he’d thought of a plan B.
His safeguard was to tuck the operative part of the coordinates into the trinomes in a form that could be recovered even without access to the phone lists. Specifically, he had embedded the last four digits of the latitude, the last three digits of the longitude, and the number of feet within the series of two-digit numbers derived from the trinomes. In other words, he’d put these digits only through the first layer of encryption, applying the two tables. He hadn’t enciphered them further using the phone lists.
The tricky thing about this “back door” was that there was no obvious marker indicating where in the message it was. Regan had embedded the information as a footnote, immediately following the word that signaled the bookending of one set of coordinates (“END” in the above example). But the only way to find it was to decrypt the entire message, and that required access to the phone lists. It was a catch-22 that defeated the purpose of a back door.
However, Olson had the advantage of knowing eleven of the twelve coordinates, which helped him figure out what the back-door digits for those sites were supposed to be in Regan’s scheme. Scanning the two-digit numbers he’d derived from the trinomes, he was able to find those embedded coordinates. That allowed him to make a reasoned guess about where the twelfth set of coordinates, in its abbreviated back-door form, was most likely to be found among the lines and lines of two-digit numbers. He could now focus his attention on a small subset of numbers in the series.
As he studied them, he came across two consecutive pairs of numbers that looked different from the rest.
(02 53) (31 27)
What was odd about the two was that the second number in each pair—53 and 27—was a lot higher than the second number in most other pairs of the series. To convert the first pair into a letter or digit according to Regan’s decipherment rules, Olson would have had to find the fifty-third character in the second listing on one of the phone lists. It was unlikely for a name, followed by an extension number, to contain such a large number of characters, Olson thought. The same was true of the second pair. And so, even though Olson didn’t have the phone lists, he deduced that the two pairs were not part of the code.
This meant that the numbers written in one string, “02533127,” had to be one of Regan’s back doors. Olson broke that down into a latitude, longitude, and number of feet and gave it to Carr.
Carr had asked an agent from the squad to stay back near Pocahontas to continue the search for the remaining package. He called him and relayed the coordinates that Olson had derived. Hours later, the agent called him back, exultant. The twelfth trove of secrets had been found.
Having confirmed the right location for the last package, Carr and his fellow investigators realized what Regan had gotten wrong when he wrote down the plaintext coordinates for that site. He had transposed a couple of digits. Once again, as with the spelling errors in his offer letter, his dyslexia was to blame.
With the Virginia packages safely back in the government’s custody, Carr turned his attention to the secrets buried in Maryland. Given the relative ease with which the FBI had found the stashes in Pocahontas, Carr was optimistic about recovering the seven packages at Patapsco Valley State Park in short order. He anticipated few challenges in decoding the encrypted coordinates for those sites—with Regan’s help—and then digging up the packages.
He couldn’t have been more wrong.
CHAPTER 10
MR. EIGHTY PERCENT
One afternoon in early April, Carr rang the doorbell of Anette Regan’s house in Bowie. He had called earlier to say that he’d be stopping by with yet another search warrant.
“What are you looking for now?” she had asked on the phone, sounding exasperated. She had expected Regan’s sentencing to mark the end of her nightmare. After eighteen months of trauma, she was desperate to piece the shambles of her life back together and find a way forward f
or herself and the kids. She would have given anything not to have another FBI agent walk through her house ever again.
Carr hadn’t told her what the search was for. She answered the door, an irritated look on her face.
“Would you like to see the search warrant?” Carr asked.
“No,” she replied.
“OK,” he said. “This will take a second.”
Stepping into the house, he walked to the bookshelves in one corner of the living room. Regan had told him exactly where to look for what he wanted. Carr bent down to access the bottom bookshelf. Four books in—just as Regan had said—was Regan’s junior high school yearbook.
Carr pulled it out and inspected it. The green hardbound cover said “1977” in a large white font, with the words “Mill Lane Junior High” printed across.
Anette watched him from where she stood in the living room. She looked mystified. “What do you want that for?” she asked.
Carr didn’t answer the question. Instead he filled out an inventory sheet indicating what he was taking, gave her a copy, and said a polite “Thanks” on his way out.
A day or so later, Carr went back to visit Regan. At the first debriefing session, Regan had not had any difficulty explaining the scheme he had used for the Virginia code. At the time, Carr had not asked him about the Maryland code because the encoded coordinates were yet to be recovered from the toothbrush holder. Now, having obtained both the code sheet and the yearbook on which the code was based, Carr asked Regan to break out the Maryland coordinates.
Regan looked intently at the sheet and went through the yearbook. Finally, after having scrutinized the materials for several minutes, he looked up at Carr apologetically.
He could not remember the key to his own code.
• • •
In his six years as a cryptanalyst at the FBI, Olson had collaborated with colleagues of many different backgrounds to break codes. Besides working with fellow cryptanalysts, he had brainstormed with field agents of the bureau, worked with jail authorities, and partnered with linguists when cracking codes written in foreign languages. What he’d never done—or thought he would never do—was work on decrypting a code in collaboration with the person who had written it.
Yet that was the task that Olson had before him when he drove up from Quantico to the Alexandria courthouse one morning in mid-April. Carr was waiting for him. The two men went down to the conference room in the basement where Carr had interviewed Regan before.
In the days prior, Olson had familiarized himself with the code. He didn’t have to look closely to see how different it was from the trinomes Regan had used to encrypt the coordinates for the sites in Virginia. The text was typewritten, unlike the Virginia “letters,” which had all been written by hand. Even though much of the code did consist of trinomes, these trinomes followed neither of the two rules obeyed by the handwritten trinomes—that of the second digit always being a number from 1 to 5, and the last digit always being 1 or 2. It was hard to find any obvious pattern in them. The series began:
146 051 400 356 370 035 739 190 454 413 . . .
Another thing that was different about the Maryland code was that the trinomes were interspersed with alphanumerical groupings containing three characters—each a two-digit number followed by an A. For example, if we pick up the series from where we left off above, at “413” and continuing on to the end of the first line, the text read:
. . . 413 958 431 13A 11A 40A 775.
At the top of the sheet, above the twelve lines of code, was an intriguing phrase—the only intelligible text on the sheet. It said: “Number One.”
Olson had also spent several hours studying the yearbook, whose pages were filled with mug shots of ninth graders who—along with Regan—had graduated from Mill Lane in 1977. The code had to be based on the pictures and names of the students. Sitting at his desk in Quantico, Olson had tried, unsuccessfully, to uncover the encryption scheme that tied the yearbook to the code sheet.
The U.S. marshals brought Regan into the room and unshackled him. He sat down at the table and nodded a greeting at Carr.
Carr introduced Olson, who was sitting across from Regan.
Regan said he remembered him from the trial, when Olson had taken the stand to testify. “Did you agree with what I said?” Olson asked jokingly. Regan didn’t react.
It was the first time that Olson was getting a close look at the man whose trinomes had given him sleepless nights. This morning, however, Olson the code breaker and Regan the code maker were no longer adversaries. In the game of hide-and-seek they had been playing since August 2001, Regan was now as much of a seeker as Olson. Their shared goal was to decipher the encrypted coordinates with the help of the yearbook. Olson’s job was to help Regan recall how the code worked.
He slid the yearbook across the table to Regan.
Regan flipped the pages, scanning the rows upon rows of classmates staring back at him, their midteen portraits reminiscent of that twilight zone between childish innocence and youthful, self-confident immaturity. In the rows he spotted the gang from his neighborhood—Brian Wagner, Bob Florio, Cliff Wagner—and friends like Peter Klopfer, who had been with Regan in remedial classes prescribed to slow learners and weathered some of the same insults as Regan. On the twenty-fifth page, at the bottom corner, was Regan himself, looking at the camera with a shy smile, a thick and shiny mop of hair covering his forehead all the way to the tops of his eyebrows.
If these pictures brought back memories of ninth grade, the notes that his classmates had written on the pages brought them into sharper focus. There was the one signed by Richard DiBernardo, thanking Regan for helping him carve a wooden eagle in their shop class. There were others that were far less kind—little reminders of the bruises that Regan’s psyche had endured during those adolescent years. “Good luck in 10th grade . . . you’ll need it,” said one.
Regan also had in front of him a copy of the code sheet.
“Let’s start right here at the top,” Olson said, pointing to the first line of text on the page. “What is ‘Number One’?”
Regan remembered that it was a reference to his own picture. He was Number One.
Olson told Regan what his best guess was for how the trinomes in the code related to the yearbook. He believed the first two digits of each trinome as well as the alphanumeric groups were linked to a picture. Was that a correct assumption? he asked Regan.
Regan confirmed that it was.
Next, Olson drew Regan’s attention to an alphanumeric group that occurred seven times through the twelve lines of code. It was 13A. Since Regan had buried seven packages, it appeared to be some kind of marker related to every package. What could it be?
Looking in the yearbook, Olson put his finger on Regan’s portrait—Number One—and counted forward to the thirteenth picture. Unlike the other mug shots, this one was a portrait of a student wearing a mask of an old man with a white Amish-style beard and bushy eyebrows. The name printed below it read “Mystery Man.”
In the white space between the bottom edge of the picture and the printed name, Regan had written “Frank.”
Olson had noticed it earlier while studying the yearbook. All the other handwriting in the book, including comments written by Regan, was in blue ink. “Frank” was the only word scribbled in black, with what Olson was certain was a gel pen. Gel pens didn’t really come to market until the 1980s, indicating that Regan couldn’t have written that word when he graduated from Mill Lane. Besides, the true identity of the student who had posed as Mystery Man was revealed elsewhere in the yearbook: it wasn’t a boy but a girl named Karen.
All of that left no doubt that Regan had labeled Mystery Man as “Frank” when he made the cipher.
What was unique about “Frank” that Regan had needed the name for the encryption? It couldn’t have been the number of letters, Olson reasoned, because there w
ere several other names in the yearbook with five letters.
If the first two digits of every trinome represented a picture, Regan could have used a maximum of ninety-nine pictures, counting forward from himself. When Olson sorted those ninety-nine names alphabetically by first initial, typing them into an Excel sheet, he saw that none of them began with an F.
Right away, it became clear why Regan had thought to write “Frank.” He needed an F—for “feet.”
With that little breakthrough, Regan remembered that all the alphanumeric groups in the code encoded letters and not numbers. Just as 13A resolved to F from “Frank,” 11A resolved to the first initial of the student whose picture was eleventh from Regan’s, and so on. When Olson translated them into letters, the majority of them were revealed to be package descriptions: CD for “Compact Discs,” SP for “Small Package,” BP for “Big Package,” and LP for “Large Package.” The letter T was interspersed throughout the message as well, representing “tree.”
The decoding of the alphabetic characters in the message confirmed that the first two digits of every trinome did indeed signify the position of portraits in the book relative to Regan. It also indicated that the remainder of the text encoded digits. Excited to be on the right track, Olson quizzed Regan about the next step in the decryption. What was the relevant information to be extracted from the picture that a trinome led to? Was it the total number of letters in the name? How was this information supposed to be combined with the trinome’s third digit?
Usually one to speak in rapid-fire sentences, Olson had to remind himself to ask the questions slowly and deliberately, as Carr had advised. “Don’t overwhelm him,” Carr had said, having noticed in prior debriefings that Regan took longer than most people to process and react to what he heard. Olson could now see what Carr meant. Through the afternoon, as he continued to probe Regan, Olson kept hitting a wall. Regan scratched his head and tried out different things with the names in an effort to remember how to move past the pictures. He couldn’t.
The Spy Who Couldn't Spell Page 21